Methods of intellectual analysis of information by virtualization and neural networks

DOI: 10.31673/2412-9070.2020.062432

Abstract

The use of information and knowledge extracted from a large amount of data benefits many applications, such as market analysis and business management. In addition, a lot of information is transmitted in graphical form. In this article, we will consider the following successful methods that can improve the efficiency of information retrieval when importing text and also when analyzing and generating images. In addition, the possibility of using computer vision in real time and comparing the performance of image classification algorithms will be investigated. A Docker container method is also proposed to reduce memory and CPU overload. The proposed method is container-based virtualization, which is performed not in the full operating system, but in private instances of the host operating system. Tests were presented to show performance differences of two popular methods of deploying the TensorFlow framework for deep learning. The methods of deploying are TensorFlow GPU from a docker container, and native source version of TensorFlow. Popular neural network models were used to test the performance, such as: ResNet 50, InceptionV3, VGG16, AlexNet, DCGAN, BigGAN and AutoGan. These findings indicate that TensorFlow GPU from a docker method is reliable and there is no lack of performance of Docker for deep learning. Furthermore, we examine the results of different types on neural network architectures, and show how the use of containerization for deep learning can be beneficial for developers and researchers.

Keywords: text analysis process; methods; computer vision; artificial neural networks; machine learning; TensorFlow; OpenCV; NAS; generative adversarial network; AutoGAN; BigGAN; CNN; ResNet 50; InceptionV3; VGG16; AlexNet; CTL-10; CIFAR-10; virtualization; virtual machine; Docker container; processor; memory; performance evaluation.

References
1. Barret Zoph, Quoc V. Le. Neural architecture search with reinforcement learning / arXiv preprint arXiv:1611.01578, 2016.
2. Advanced Memory Reusing Mechanism for Virtual Machines in Cloud Computing. Gursharan Singh’ Sunny Behal’ Monal Taneja 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015).
3. Salton G., Buckley C. Term-Weighting Approaches in Automatic Text Retrieval, Information Processing and Management // An Int’l J. 2018.
4. Applying Data Mining Techniques for Descriptive Phrase Extraction in Digital Document Collections / H. Ahonen, O. Heinonen, M. Klemettinen, A. I. Verkamo. 2015. Р. 2–11.
5. Wu S.-T., Li Y., Xu Y. Deploying Approaches for Pattern Refinement in Text Mining // Proc. IEEE Sixth Int’l Conf. Data Mining (ICDM ’06). 2016. Р. 1157–1161.
6. The OpenCV Tutorials [Електронний ресурс]. URL: docs.opencv.org/2.4/opencv_tutorials.pdf.
7. Efficiency analysis of provisioning Microservices / H. Khazaei, C. Barna, N. Beigi-Mohammadi, M. Litoiu // School of Computer Science, York University Toronto, Ontario, Canada.
8. Krizhevsky A., Sutskever I., Hinton G. E. Imagenet classification with deep convolutional neural networks. 2012. Р. 1097–1105.
9. Chen L.-C., Collins M., Zhu Y. Searching for efficient multiscale architectures for dense image prediction.
10. Liu C., Chen L.-C., Schroff F. Auto-deeplab: Hierarchical neural architecture search for semantic image segmentation.
11. Amc: Automl for model compression and acceleration on mobile devices / Yihui He, Ji Lin, Zhijian Liu [et al.] // In European Conference on Computer Vision, Springer, 2018. Р. 815–832.
12. Designing neural network architectures using reinforcement learning / Bowen Baker, Otkrist Gupta, Nikhil Naik, Ramesh Raskar. 2016.
13. Efficient Neural Architecture Search via Parameter Sharing / H. Pham, M. Y. Guan, B. Zoph, Q. V. Le. 2018.
14. Practical blockwise neural network architecture generation / Zhao Zhong, Junjie Yan, Wei Wu [et al.]. 2018. P. 2423–2432.
15. Learning Transferable Architectures for Scalable Image Recognition / Barret Zoph, Vijay Vasudevan, Jonathon Shlens, Quoc V. Le. CVPR, 2018.
16. Xie L., Yuille A. Genetic cnn // IEEE International Conference on Computer Vision (ICCV). 2017. Р. 1388–1397.
17. Jin H., Song Q., Hu X. Auto-keras: Efficient neural architecture search with network morphism, 2018.
18. Liu H., Simonyan K., Yang Y. Darts: Differentiable architecture search, 2018.
19. Ahmed K., Torresani L. MaskConnect — Connectivity Learning by Gradient Descent. ECCV, cs. CV, 2018.
20. Chen L.-C., Papandreou G. Rethinking atrous convolution for semantic image segmentation, 2017.
21. Wang Z., Chang S., Yang Y. Studying very low resolution recognition using deep networks. 2016. Р. 4792–4800.
22. Chrabaszcz P., Loshchilov I., Hutter F. A downsampled variant of imagenet as an alternative to the cifar datasets: arXiv preprint arXiv: 1707.08819, 2017.
23. Liu C., Zoph B., Neumann M. Progressive Neural Architecture Search. 2018. P. 19–34.